Python implementation of normalised pairwise variability index (nPVI), as defined in Grabe & Low (2002)
The normalised pairwise variability index (nPVI) is a measure of the average variation of a set of durations, ordered by successive pairs. First conceived for the rhythmic classification of speech by Grabe & Low (2002), nPVI has since also been utilised in music perception and cognition research, e.g. Toussaint (2012).
While a previous implementation of nPVI does exist for MATLAB (Livingstone 2011), Python is more commonly used by researchers in the above fields. The implementation provided here is incredibly simple and does not require the installation of any additional packages beyond those within the standard Python 3+ library.
TODO
from npyvi import npvi
# Calculating nPVI for a single array
>>> npvi(1, 2, 3)
53.333333333333336
# Calculating nPVI seperately for multiple arrays
>>> npvi([1, 2, 3], [3, 3, 6], [4, 6, 8])
[53.333333333333336, 33.33333333333333, 34.285714285714285]
From the root directory, run:
python -m unittest
- Grabe, E., & Low, E. L. (2002). Durational variability in speech and the rhythm class hypothesis. In C. Gussenhoven & N. Warner (Eds.), Papers in laboratory psychology (pp. 515–546). Cambridge University Press.
- Livingstone, L. (2011). Normalized Pairwise Variability Index MATLAB Central File Exchange. Retrieved October 30, 2022.
- Toussaint, G. T. (2012). The Pairwise Variability Index as a Tool in Musical Rhythm Analysis. In E. Cambouropoulos, C. Tsougras, P. Mavromatis, & K. Pastiadis (Eds.), Proceedings of the 12th International Conference on Music Perception and Cognition and the 8th Triennial Conference of the European Society for the Cognitive Sciences of Music (pp. 1001–1008).
Huw Cheston – @HuwCheston – [email protected]
Distributed under the MIT license. See LICENSE for more information.